Direct Inference of Cell Positions using Lens-Free Microscopy and Deep Learning

Philipp Gruening, Falk Nette, Noah Heldt, Ana Cristina Guerra de Souza, Erhardt Barth

Keywords: semantic segmentation, deep learning, in-line holography

TL;DR: Estimating the positions of cells with CNN-based segmentation on raw holographic images is faster and better than an image reconstruction approach.

Abstract: With in-line holography, it is possible to record biological cells over time in a three-dimensional hydrogel without the need for staining, providing the capability of observing cell behavior in a minimally invasive manner. However, this setup currently requires computationally intensive image-reconstruction algorithms to determine the required cell statistics. In this work, we directly extract cell positions from the holographic data by using deep neural networks and thus avoid several reconstruction steps. We show that our method is capable of substantially decreasing the time needed to extract information from the raw data without loss in quality.

Registration: I acknowledge that publication of this at MIDL and in the proceedings requires at least one of the authors to register and present the work during the conference.

Authorship: I confirm that I am the author of this work and that it has not been submitted to another publication before.

Paper Type: both

Primary Subject Area: Segmentation

Secondary Subject Area: Application: Other